DOCUMENT RESUME
ED 348 647
CS 010 992
AUTHOR
TITLE
PUB DATE
NOTE
PUB TYPE
Foorman, Barbara R. ; And Others
Links among Segmenting, Spelling, and Reading Words
in First and Second Grades.
Apr 92
20p.; Paper presented at the Annual Meeting of the
American Educational Research Association (San
Francisco, CA, April 20-24, 1992).
Speeches/Conference Papers (150) — Reports -
Research/Technical (143)
EDRS PRICE
DESCRIPTORS
IDENTIFIERS
MF01/PC01 Plus Postage.
Analysis of Variance; "Beginning Reading; Grade 1;
Grade 2; Phonology; Primary Education; Reading
Research; "Spelling
Houston Independent School District TX; "Segmentation
Skills
ABSTRACT
Twenty first graders and 20 second graders in
Houston, Texas, were examined on skills in segmenting, reading, and
spelling 50 words with regular and exceptional spelling patterns. By
using the same words for each task, it was possible to assess the
interrelationships among these skills on a word by word, child by
child basis. A multivariate analysis of variance was conducted on
mean difference scores for segmentation-reading,
segmentation-spelling, and reading-spelling. Positive differences
between measures were observed, except in the segmentation-reading
analysis of regular words requiring the deletion of consonant blends
and medial sounds. In addition, graphical analyses showed a greater
probability of correct reading and spelling given correct
segmentation than incorrect segmentation. Results were interpreted to
support a computational notion of phonology as a prerequisite to
reading and spelling, with a mere reflective notion explaining the
reciprocal relation between reading and segmentation of complex
spelling patterns. (Four tables and two figures of data are included;
21 references and a list of the stimuli for segmentation, reading,
and spelling tests are attached.) (Author/RS)
**************************************^
* Reproductions supplied by EDRS are the best that can be made
* from the original document.
Links in First and Second Grades
1
Links among Segmenting, Spelling, and Reading
Words in First and Second Grades
Barbara R. Foorman
Lauren Jenkins
David Francis
~-\ University of Houston
The authors would like to thank the principal, teachers, and children of St.
Vincent de Paul School, Houston, Texas for their cooperation and support. This
manuscript was presented to the American Educational Research Association
meeting in San Francisco, April 20-24, 1992. Requests for reprints should be sent to
: Barbara Foorman, Dept. of Educational Psychology, College of Education,
University of Houston, Houston, TX 77204-5874.
Running head: LINKS IN FIRST AND SECOND GRADES
o
9
ERIC
Abstract
Twenty first graders and twenty second graders were examined on skills in
segmenting, reading, and spelling 50 words with regular and exceptional spelling
patterns. By using the same words for each task, it was possibloe to assess the
interrelationships among these skills on a word by word, child by child basis- A
multivariatew analysis of variance was conducted on mean difference scores for
segmentation-reading, segmentation-spelling, and reading-spelling. Positive
differences between measures were observed, except in the segmentation-reading
analysis of regular words requiring the deletion of consonant blends and medial
sounds. In addition, graphical analyses showed a greater probability of correct
reading and spelling given correct segmentation than incorrect segmentation.
Results were interpreted to support a computational notion of phonology as a
prerequisite to reading and spelling, with a more reflective notion explaining the
reciprocal relation between reading and segmentation of complex spelling patterns.
"PERM.SSION TO REPRODUCE THIS
MATERIAL HAS BEEN GRANTED BY
TO THE EDUCATIONAL RESOURCES
INFORMATION CENTER (ERIC) "
U.S. DEPARTMENT OF EDUCATION
Office of Educational Research and improvement
EDUCATIONAL RESOURCES INFORMATION
CENTER (ERIC)
I^This document has been reproduced as
' received from the person or organization
originating
O Minor changes have been made to improve
reproduction quality
• Points ot view or opinions stated in this docu-
ment do not necessarily represent official
OERI poailion or policy
Links in First and Second Grades
2
It is no longer surprising to find a significant connection between children's
awareness of phonological units in words and their ability to read and spell (see
Adams, 1990 and Goswami & Bryant, 1990 for reviews). In fact, there are
longitudinal and training studies that prove a causal connection between
phonological awareness and success in reading and spelling (Bradley & Bryant,
1985: Lundberg, Frost, & Petersen, 1988). Moreover, there is converging evidence
that it is a deficit in phonological skills rather than a lag in the development of
phonological skills tnat differentiates dyslexic from poor readers (Rack, Snowling, &
Olson, 1992; Francis, Shaywitz, Stuebing, Shaywitz, & Fletcher. 1992). However,
before we rush to salvage the fiiture of children's literacy with phonological
awareness programs, we need to know more precisely which phonological unit(s)
children need to be aware of-syllable, phoneme, or something in between called
onset-rime (Treiman, 1985)--and exactly how awareness facilitates acquisition of
reading and spelling. Accordingly, in the present investigation we asked first and
second graders to segment, read, and spell the same fifty words so that on a child by
child, word by word, basis we could address how ability to segment the sounds in
particular words related to the way a child read and spelled those same words.
In our previous research (Foorman, Francis, Novy, & Liberman, 1991;
Foorman & Francis, 1992; Foorman & Liberman, 1989), we used Rosner and
Simon's (1971) Test of Auditory Analysis Skills(TAAS) to measure segmentation skill
and our own experimental word list to measure reading and spelling skills in six
classrooms of first graders receiving different amounts of letter-sound instruction.
Although classrooms did not differ in TAAS scores collected in October, classrooms
with more letter-sound instruction improved at a faster rate in correct spellings and
readings. Individual growth models analysis indicated that segmentation scores
obtained in October predicted overall performance in reading and spelling. Growth
in segmentation scores predicted overall performance in spelling but only predicted
end-of-year differences in regular and exception-word reading. Finally, better
reading of regular words in October was associates with faster growth in spelling,
and better spelling of regular words in October was predictive of May word reading,
after conrtrolling for the facilitative effect of more letter-sound instruction on
growth rate.
We interpreted these results as suggestive of a uni-directional link between
segmentation and reading, in support of Bradley and Bryant (1985), and a bi-
directional link between segmentation and spelling and between spelling and
reading. But one concern we have about generalizing these results (and others in the
literature) is that the linguistic effects on segmenting the words on the TAAS are
bound to be different from the linguistic effects on reading and spelling words from
our experimental list simply because the TAAS consists ofdifferent words. However,
in the literature it is common practice to use different lists of words to assess
phonological awareness, reading, and spelling.
Rosner and Simon's (19vl) TAAS, based on the work of Bruce (1964), is not
in fact a test of auditory analysis skills. Rather, it is a phonological test of skill in
segmenting, deleting, and eliding initial, medial, and final sounds in words at the
syllable and phoneme level. A sample item is "Say carpet, now say it again without
car" {pet). Other items are the following, with the sound to be deleted noted
parenthetically: bel(t) and ti(me); (l)end and (s)mile; cr(e)ate and cont(in)ent. In each
case the elided sounds result in a real English word. Rosner and Simon (1971) gave
the TAAS to children whose ages ranged from five years to eleven years and found
that number correct on the TAAS correlated with reading achievement. They also
found that five and six year olds were better at deleting final sounds than initial
sounds, a finding supported by Content, Kolinksy, Morais and Bertelson (1986) on a
slightly different deletion task* Finally, Rosner and Simon confirmed Bruce's finding
Links in First and Second Grades
3
that deletion of medial sounds (as second consonant in a consonant blend or middle
syllable in a three-syllable word) was difficult even for the oldest children.
Critics of deletion tasks such as Rosner and Simon's claim that a child
already must be able to read and spell to perform well on the task (see Adams,
1990). This may very well be true for deleting medial sounds. However, deletion of
final and initial sounds may tap into computational knowledge of phonology rather
than reflective knowledge of phonology (rerfetti, 1991). This computational
knowledge may manifest itself as the preliterate skill that facilitates acquisition of
literacy. From Goswami and Bryant's (1990) perspective, we are not speaking of
deleting initial and final phonemes. Rather, we are speaking of deleting beginning
and ending sounds that coincide with Treiman's (1985) intrasyllabic units or onset
and rime. Indeed, on the TAAS, deletion of the initial consonant in a single-syllable
word (e.g., (g)ate) is easier than splitting apart the onset (e.g., (t)rail or g(l)ow).
However, the TAAS has no items representing rime deletion. Rime deletion is
typically tapped through Bradley and Bryant's (1985) oddity task where subjects are
asked to fina the odd word in a sequence of words such as "hop," "rail," and "mop."
Kirtley, Bryant, MacLean and Bradley (1989) found that it was much easier for five
year olds to find "rail" in the above example than to find the odd word in the
sequence "mop," "whip," and "lead." In other words, it is easier to delete rimes than
final phonemes. However, the fact remains that young children can delete the final
sounds on the 7>L45-sounds that coincide with phonemes. Perhaps, as Goswami
and Bryant (1990) suggest, they are able to do so simply by learning to "drop the end
bit" of the word, a strategy that relies on crude computational knowledge of sub-
syllablicparts rather than reflective knowledge of the abstract category of phoneme.
Tne objective of the present investigation, then, was to revise the TAAS to
include words from our experimental word list-words with predictable and less
predictable spelling patterns, called "regular" and "exceptional" here, although the
variable is clearly continuous, rather than dichotomous (see Goswami & Bryant,
1990, pp. 39-42). Consistent with the view that segmentation skill is a prerequisite to
reading and spelling performance, we hypothesized that for first and second grade
children, segmentation performance would exceed reading and spelling
performance. Accordingly, we also hypothesized that reading and spelling ; esponses
would be more advanced for words that were segmented correctly.
Method
Subjects
Forty boys and girls, half from grade 1 and half from grade 2, were selected
on the basis of parental consent and Stanford Achievement Test . Form J, scores
within six montns of grade level Because testing occurred at the end of the year,
effort was made to limit selection to grade equivalent scores ranging from 1.6 to 2.6
for first graders, and 2.6 to 3.6 for second graders. However, the generally high
reading level of students necessitated acceptance of 2 subjects in first grade and 11
subjects in second grade with scores above this range. One subject selected in
second grade scored below this range. Consequently, average reading achievement
scores were 2.2 in first grade and 3.6 in second grade. Subjects attended a parochial
school in southwest Houston, composed of cildren from the surrounding middle
class neighborhood. Seventy percent of the subjects were white, 17.5% were
Hispanic, and 12.5% were black.
Materials
Fifty real English words were selected for the reading, spelling, and
phonological segmentation measures: 58% were single-syllable words, 28% were
two-syllable words, and 14% were three-syllable words. Of these words, 50% had
regular spelling-to-sound correspondences, such as link, and 50% had exceptional
correspondences, such as climb. The words were further categorized by number of
4
Links in First and Second Grades
4
letters: single-syllable words consisted of 4-6 letters, two-syllable words consisted of
6-10 letters, and three-syllable words consisted of 6-10 letters.
Words were selected to include items appropriate for each grade level, based
upon cumulative vocabulary lists for grades 1-5 in the Harcourt Brace Jovanovitch
basal reading series. Items were chosen that would have a high probability of being
within the spoken, but not the written vocabulary of children at each grade level.
Thus, in selecting words appropriate for each grade level, words presented in the
cumulative vocabulary list for that level were not included, and an attempt was
made to include words presented in the vocabulary list for the next grade. Words
not presented in the cumulative vocabulary lists were determined to have
comparable levels of frequency of usage, as assessed by Carroll, Davies, and
Richman (1971).
In addition, words were selected on the basis of Rosner and Simon's (1971)
seven categories, with regular and exception words equally represented in each
category: 1) omission of the final syllable of a two-syllable word; 2) omission of the
initial syllable of a two-syllable word; 3) omission oi the final consonant of a one-
syllable word; 4) omission of the initial consonant of a one-syllable word; 5)
omission of the first consonant of a consonant blend; 6) omission of a medial
consonant of a consonant blend; 7) omission of a medial syllable. For purposes of
analysis, these seven categories were collapsed into three categories of initial,
medial, and final. (A copy of the phonological segmentation test is included in the
Appendix.)
Procedure
First and second grade chidlren were group tested in the classroom by their
respective teachers on their ability to spell 25 regular and 25 exception words. The
spelling test was administered over the period of a week, witt 1 0 words presented
each day. Using a pencil and a piece of lined paper, numbered 1-10, the children
were instructed to listen to each sound and attempt to write the letters on the paper
even if they did not know how to spell a word. Each word was presented singly, as
well as embedded in a sentence.
During one 30-minute individual session, each child was asked to segment
each word by deleting initial, medial, or final sounds at the syllabic or phonemic
levels. To demonstrate the segmentation task, the examiner showed the child an 8.5
in, x 11 in. card on which pictures of a cow and a boy's head were drawn side by side.
The examiner instructed the child to say "cowboy." After the child responded, the
examiner covered the picture of the boy and asked the child to say "cowboy" again,
but without "boy." If the response was correct, "cow," the next demonstration item,
"toothbrush," was presented using puppets, with one puppet teaching another puppet
how to segment the word with the child's assistance. The same procedure was
repeated for the third demonstration item, "bake." If the child failed a
demonstration item, an attempt was made to teach the task by repeating the
demonstration procedures. In the present study, all subjects completed the
demonstration items and proceeded to the test. The experimenter pronounced the
specific sound(s), not the letter name, to be omitted. If the child failed to respond to
an item, it was repeated exactly as it was first stated. If the child again did not
respond, an item score of zero was recorded and the next item was presented. All
test items were presented.
After first and second grade children completed the segmentation task, they
were shown each of the 50 words on a series of 3 in. x 5 in. cards, and asked to try to
read each word. Responses to the segmentation and word reading tasks were
audiotaped. Segmentation, reading, and spelling responses were scored 1 if correct
and 0 if incorrect. Interrater reliability was virtually 100%.
Results 3
Links in First and Second Grades
5
Means and standard deviations for segmentation, reading, and spelling scores
are provided in Table 1. As expected, second grade scores were higher than first
graae scores in reading and spelling, F(l,38) = 14.76 and 16,49,/? < .001,
respectively. Additionally, all subjects read and spelled the 25 regular words more
accurately than the 25 exception words, F(l,38) = 12.31,/? < .01 and 22.73,/? <
.001, respectively. In fact, second grade performance was near ceiling--95%~on
reading aioud regular words.
Insert Table 1 about here
Analyses of Variance
Segmentation means were analyzed for effects of grade, category, and
wordtype. There were significant interactions of grade x category and category x
wordtype, F(2, 37) = 3.88, p < .05 and 16.32, /? < .001, respectively. Post hoc
analysis, utilizing the Bonferoni adjustment for alpha (.05/2 = .025), revealed grade
level differences in segmenting final compared to medial sounds, F(l, 38) = 6.43,/?
< .025. As can be seen in Table 2, both grades were near ceiling on segmenting final
sounds, with means ranging between 0 and 1. However, in segmenting medial
sounds, first graders' performance was noticeably lower than second graders' (i.e.,
.176 vs. .424). Likewise, in the category x wordtype interaction, it was the contrast
between medial and final sounds that was significant, F(l, 38) = 8.83,/? < .01. From
Table 2 we again note the near ceiling performance on deleting final sounds across
type of word. The source of significance lies in the relative inability to delete medial
sounds in exception words compared to regular words (i.e., .253 vs. .346). This
difference suggests that subjects abandoned a phonological strategy and adopted a
visual-orthographic strategy when deleting medial sounds. However, the visual-
orthographic strategy often failed on words with less predictable spelling patterns,
that is, exception words.
Insert Table 2 about here
From the means of Table 2 it appears that initial deletion was more difficult
than final deletion. However, it should oe pointed out that within this category
performance differed for one-syllable words requiring deletion of the initial
consonant in a consonant blend (e.g., (b)rag) as opposed to the initial consonant
alone (e.g., (l)ink). A frequency analysis revealed that, on average, only 58% of
subjects correctly deleted the initial consonant when it was part of a blend in
contrast to 92% of the subjects when it was not part of a blend.
To assess the hypothesis that segmentation performance would exceed
reading and spelling performance, and to examine the influence of category,
wordtype and grade on the differential performance on these measures, each child's
average difference score was computed for segmentation and reading, segmentation
and spelling, and reading and spelling, on a word by word basis. Because responses
on each variable were coded 0 tor incorrect and 1 for correct, difference scores
between any two variables could range from -1 to + 1. A difference score of 0 would
indicate that performance on the two variables was comparable. For example, a
positive difference score in the analysis of segmentation-reading would indicate that,
on average, segmentation exceeded reading; a negative difference score would
indicate the converse. For each of the overall average difference scores, a
multivariate analysis of variance with repeated measures was conducted, with grade
as the between subjects factor and wordtype (regular or exception) and category
(deletion of initial, medial, or final sounds) as within subjects factors. Because
ERLC
Links in First and Second Grades
6
category is relevant only to segmentation, it was not included as a factor in the
reading-spelling analysis.
A summary of the multivariate analyses of variance with repeated measures
for each pair of variables is presented in Table 3, Inspection of Table 3 reveals no
significant grade level differences in overall mean difference scores for
segmentation and reading, segmentation and spelling, or reading and spelling.
However, there were significant interactions ot grade x category and wordtype x
category in the analyses involving segmentation.
Insert Table 3 about here
Observation of positive mean difference scores in Table 4 indicates that, on
average, segmentation was superior to reading with exception words and with
regular words requiring the deletion of final sounds. Negative mean difference
scores were observed only with regular words involving deletion of initial and
medial sounds. In the analysis of difference scores, there were significant effects of
wordtype, F(l f 38) = 38.75, p < .001 and category, F(2, 37) = 5.13, p < .01, as well
as a significant wordtype x category interaction, F(2, 37) = 4.92, p < .01. Post hoc
analysis of this interaction, with Bonferoni adjustment tor alpha (.05/2 = .025),
revealed a differential effect of wordtype in the contrast of mean difference scores
for words involving medial and final sounds, F(l, 38) = 8.81,/? < .025. Specifically,
differences between exception and regular words were maximized in deletion of
medial sounds and minimized in deletion of final sounds.
Insert Table 4 about here
Positive mean difference scores were also observed in the segmentation-
spelling analysis for words of both wordtypes, particularly exception words, with
mean difference scores of .190 for regular words and .357 for exception words. The
wordtype effect was significant, F(l, 38) = 7.03,/? < .01. Although a significant
category x grade interaction was observed, F(2, 37) = 4.54,/? < .05, post hoc analysis
indicated no significant grade level differences between the contrasts of interest:
initial versus final; medial versus final; and initial versus medial sound deletion. The
significant effect of category, F(2, 37) = 13.32, p < .001, was due to the fact that the
segmentation-spelling difference was greater (i.e., mean of .478) for words involving
final sound deletion in contrast to medial sound deletion (i.e., mean of .047).
In addition, positive mean difference scores were noted in the reading-
spelling analysis of words of both wordtypes. There were no significant effects of
wordtype or grade (p > .05).
Graphical Analyses
To examine the hypothesis that reading and spelling responses would be
more advanced for children who could segment the words, Figure 1 (exception
words) and Figure 2 (regular words) depict the conditional probability of correctly
reading and spelling a particular word given correct or incorrect segmentation of
that same word for grades 1 and 2. For example, in the upper lefthand plot of Figure
1, the conditional probability of correct spelling (x axis) given incorrect
segmentation is plotted against the conditional probability of correct reading (y axis)
given incorrect segmentation of exception words. Squares represent first graders and
pluses represent second graders. Some random noise has been introduced to reduce
overplotting, a technique known as jittering (Chambers, Cleveland, Kleiner &
Tukey, 1983).
Insert Figures 1 and 2
Links in First and Second Grades
7
These plots reveal that when correct reading .and spelling is considered, a
similar pattern is evident regardless of segmentation response to that word. In all
eight plots of Figures 1 and 2, if an imaginary line is drawn at a 45° angle between
the x and y axes, the clustering of symbols above the line indicates that there is a
greater probability of correct reading than correct spelling given either correct or
incorrect segmentation for grades 1 and 2. In addition, regardless of segmentation
response, there is a greater probability of correct reading and spelling performance
for second graders than first graders, reflected by the slightly greater incidence of
symbols for grade 2 above a probability value of .50 on the x and y axes given correct
or incorrect segmentation. Importantly, there is a greater probability of correct
reading and spelling given correct segmentation, as indicated by the slightly greater
incidence of symbols above a probability value of .50 along the x and y axes for
correct segmentation than incorrect segmentation.
Discussion
To examine the relations among phonological segmentation, reading, and
spelling, two hypotheses were assessed:
1. Phonological segmentation would exceed reading and spelling
performance.
2. Reading and spelling performance would be more advanced for words
segmented correctly.
First and second graders were asked to segment, read, and spell the same 50
words so that analyses could be conducted on a word by word, child by child basis.
Two types of analyses were performed-analyses of mean difference scores by grade
for segmentation, reading, and spelling and graphical analyses of the conditional
probabilities of correct or incorrect performance among these measures.
Analyses of mean difference scores for segmentation, reading, and spelling
revealed no significant differences between grades. There were, however, significant
differences between type of word and category of segmentation. The wordtype effect
was due to the relative ease of segmenting, reading, or spelling words with regular
spelling-to-sound correspondences, such as link, compared to words with exceptional
correspondences, such as climb.
The category effect in the analyses concerning segmentation was explained
by the fact that differences were maximized when final sounds were deleted and
minimized when medial sounds were deleted. These category effects support the
difficulty of medial sound deletion (Bruce, 1964; Rosner & Simon, 1970). In this
modification of Rosner and Simon's (1970^ Test of Auditory Analysis Skills , medial
sound deletion involved deleting the middle syllable in a three-syllable word (e.g.,
ca{na)ry) or deleting the second phoneme in a consonant blend (e.g., s(p)ite). In
contrast, final sound deletion involved deleting a final syllable in a two-syllable word
(e.g., cap(tain)) or deleting the final phoneme in a word (e.g., bor(n). The advantage
of having an orthographic image to work off of when deleting medial sounds is
obvious. In other words, prior ablity to read or spell a word is an enormous
advantage when it comes time to aurally segment medial sounds in that word.
In these results deletion of initial sounds fell mid-way in difficulty between
deletion of final sounds and deletion of medial sounds. Nearly all subjects deleted
initial consonants in a single syllable word correctly. However, just a little over half
the children in each grade deleted the initial consonant of a blend correctly. Again,
prior ability to read or spell a word would provide a usefol orthographic image as
the child tried to segment the blend in a spoken word. It is informative that when
these first and second graders made errors in deleting the initial or second
consonant in a blend, they tended to produce the rime of the word. For example,
when asked to say "crib" without the /k/, they tended to say "ib." When asked to say
"swing" without the /w/, they tended to say "mg." These kinds of errors lend support
to Goswami and Bryant's (1990) claim that beginning readers do as well as they do
Links in First and Second Grades
8
on Rosner and Simon's (1971) test because they adopt a strategy of dropping either
the initial or ending sounds. These units correspond to Treiman's (1985)
intrasyUabic divisions of onset and rime, and are for Goswami and Bryant (1990) the
orthographic/phonological units most predictive of success in reading and spelling.
Finally, the hypothesis that reading and spelling responses would be more
advanced for words segmented correctly was explored. It was found that, indeed,
when words were segmented correctly, there was a greater probability that they
would be read and spelled correctly than incorrectly, and the cojiverse was apparent,
as well.
In conclusion, the link between the way beginning readers segment a word
and the way they read or spell that word is clearly dependent upon the category of
sound to be deleted and the adequacy of their orthographic representation. Results
of this investigation are consistent with the view that awareness of beginning and
ending sounds is a prerequisite to reading and spelling, but that awareness of medial
sounds and consonant blends is a consequence of reading and spelling. We use
"beginning sound" and "ending sound" advisedly because it is not clear to us that
these children have phonemic awareness. They appear to have a sub-syllabic
awareness akin to, it not the same as, Treiman's (1985) notion of onset-rime. The
initial consonant deletions in this study are, in fact, onsets and, although children
were not asked to delete rimes, the errors they produced on deletion of consonant
blends were rimes. Therefore, we conclude that the phonological activities
most facilitative to beginning reading and spelling would be at the level of onset and
rime. Such activities might consist of alliterative and rhyming games. Indeed, Bryant
and his colleagues (MacLean, Bryant & Bradley, 1987; Bryant, Bradley, MacLean &
Crossland, 1989) found that three year olds' knowledge of nursery rhymes was
related to their sensitivity to rhyme two years later. However, we do not exclude the
value of phonological activites involving deletion of final phonemes. Clearly
beginning readers can segment and delete final sounds that are, in fact, phonemes.
And clearly they can do so for the beginning sounds-onsets-that are phonemes as
well. It may be that it is this crude computational, rather than reflective, knowledge
of phonemes that fuels the bi-directional link between phonology and spelling, a link
that supports the development of an alphabetic strategy for reading (see Frith,
1985).
Links in First and Second Grades
9
References
Adams, M.J, (1990). Beginning to read: Thinking and learning about print.
Cambridge, MA: MIT Press.
Bruce, L.J. (1964). The analysis of word sounds. British Journal of Educational
Psychology, 34, 158-170.
Bradley, L., & Bryant, P.E. (1985). Rhyme and reason in reading and spelling.
International Academy for Research in Learning Disabilities Monograph Series,
1. Ann Arbor: University of Michigan Press.
Bryant, P.E., MacLean, M., & Bradley, L, & Grassland, D. (1989). Nursery rhymes,
phonological skills and reading. Journal of Child Language, 16, 407-428.
Carroll, J., Davies, P., & Richman, B. (1971). Tfie American Heritage word frequency
book. Boston, MA: Houghton-Mifflin.
Chambers, J.M., Cleveland, W.S., Kleiner, B.K., Tukey, P.A. (1983). Graphical
methods for data analysis. Pacific Groves, CA: Wadsworth & Brooks/Cole,
Content, A, Kolinsky, R., Morais, J., & Bertelson, P. (1986), Phonetic segmentation
in pre-readers: Effect of corrective information. Journal of Experimental
Child Psychology, 42, 49-72.
Foorman, B.R., & Francis, D.J. (1992) Exploring the links among reading spelling
and phonemic segmentation during first grade. Unpublished manuscript,
Foorman, B.R., Francis, DJ., Novy, D.M., & Liberman, D. (1991). How letter-sound
instruction mediates progress in first grae reading and spelling. Journal of
Educational Psychology, 83, 456-469.
Foorman, B.R., & Liberman, D. (1989). Visual and phonological processing of
words: A comparison of good and poor readers. Journal of Learning
Disabilities, 22, 349-355.
Francis, D.J., Shaywitz, S.E., Stuebing, K.K., Shaywitz, B.A, & Fletcher, J.M. (1992).
The measurement of change: Assessing behavior over time and within a
developmental context. Manuscript submitted for publication.
Frith, U. (1988). Beneath the surface of developmental dyslexia. Surface dyslexia.
Hillsdale, NJ: Erlbaum.
Goswami, U., & Bryant, P.E. (1990). Phonological skills and learning to read.
Hillsdale, NJ: Erlbaum.
Harcourt Brace Jovanovitch. (1987). HBJ reading. Orlando, FL: Author.
Kirtley, C, Bryant, P., MacLean, M., & Bradley, L. (1989). Rhyme, rime and the
onset of reading. Journal of Experimental Child Psychology, 48, 224-245.
Lundberg, L, Frost, J., & Petersen, O. (1988). Effects of an extensive program for
stimulating phonological awareness in preschool children. Reading Research
Quarterly, 23, 263-284.
MacLean, M., Bryant, P.E., & Bradley, L. (1987). Rhymes, nursery rhymes and
reaing in early childhood. Merrillj-Palmer Quarterly, 33, 255-282.
Perfetti, C. (1991). Representations and awareness in the acquisition of reading
competence. In L. Rieben & C. Perfetti (Eds.), Learning to read (pp. 33-43).
Hillsdale, NJ: Erlbaum.
Rack, J.P., Snowling, M.J., & Olson, R.K. (1992). The nonword reading deficit in
developmental dyslexia: A review. Reading Research Quarterly, 27, 28-53.
Rosner, J. & Simon, D. (1971). The Auditory Analysis Test: An initial report.
Journal of Learning Disabilities, 4, 384-392.
Treiman, R. (1985). Onsets and rimes as units of spoken syllables: Evidence from
children. Journal of Experimental Child Psychology, 39, 161-181.
Links in First and Second Grades
10
1.
2.
3.
4.
5.
6.
7.
8.
9.
22.
23.
24.
25.
Appendix
Stimuli for Segmentation. Reading, and Spelling Tests
(Segment of word to be deleted is within pr rentheses)
(p)rey
(b)reak
(ch)rome
(b) rag
(f) lock
(drib
(g) lobe
(c) limb
(sh)oes
(sp)read
b(l)ank
s(p)ite
s(w)ing
wor(l)d
ph(r)ase
pflteid
gmeat
ca(na)ry
fan(ta)sy
hos(pi)tal
col(on)ies
no(vel)ty
worr(ied)ly
bur(gun)dy
per(son)al
(yel)low
26.
(pea)nut
27.
(tip)toe
28.
(see)saw
29.
(work)men
30.
(no)thing
31.
(some)thing
32.
star(fish)
33.
mail(box)
34.
sun(burn)
35.
be(come5
36.
for(ward)
37.
cap(tain)
38.
hand(some)
39.
fir(m)
40.
bor(n)
41.
li(me)
42.
buil(d)
43.
bow(l)
44.
bur(y)
45.
swor(d)
46.
Mold
47.
(d)ae
48.
(l)ink
49.
(m)eet
50.
30
Table 1
Mean Number of Words Correct for Grades 1 and 2
Measure
Phonological
Grade
Segmentation
Reading
Spelling
Regular Words
1
M
1 6.05
20.65
10.10
Mean
percentage
C A A r»
o4 . UO
83.00
40.00
SD
03.68
03.01
03.88
2
M
18.65
23.85
14.85
Mean
percentage
75.00
95.00
60.00
SD
03.20
02. 01
05. 1 9
Exception Words
1
M
14.40
17.75
04.1 0
Mean
percentage
58.00
71 .00
16.00
SD
03.66
03.86
03.1 3
2
M
17.35
21 .60
09.80
Mean
percentage
70.00
86.00
39.00
SD
02.70
03.12
04.79
Table 2
Segmentation Means and Standard Deviations as
a Function of Grade, Category, and Word Type
Category
Initial Medial Final
Grade
Grade 1
Mean .727 .176 .888
SD .185 .169 .223
Grade 2
Mean .776 .424 .954
SD .188 .224 .061
Word Type
Regular
Mean .783 .346 *21
SD .185 .267 .^85
Exception
Mean .719 .253 S22
SD .219 .227 .x59
32
Table 3
Multivariate Analysis of Variance with Repeated Measures
Phonemic Phonemic
Segmentation- Segmentation- Reading-
Reading Spelling • Spelling
Source
df
F
df
F
df
F
Between Subjects
Grade
1 , 38
1 .41
1 , 38
2.47
1 , 38
1 .73
Within Subjects
Word type
1 , 38
38.75***
1 , 38
7.03**
1 , 38
3.92
Category
2,37
5.13**
2,37
13.32***
2, 37
N/A
Wordtype x Grade
1-, 38
1.10
1 , 38
0.56
1 , 38
0.26
Category x Grade
2, 37
1 .66
2, 37
4.54*
2, 37
N/A
Wordtype x
Category
2, 37
4.92**
2, 37
1 .37
2,37
N/A
Wordtype x
Category x Grade
2, 37
0.32
2, 37
0.06
2, 37
N/A
*p_<.05
**£< .01
***£<. 001
1 *
ERIC
Table 4
Mean Difference Scores for Phonological Segmentation and
Reading with Words Grouped According to Wordtype and Category
Wordtype
Regular Exception
Category Mean Mean Mean
Total
Initial -0.110 0.228 0.059
Medial -0.450 0.343 -0.054
Final 0.004 0.187 0.100
Mean
Total -0.185 0.253
Note. Positive mean difference scores indicate the extent to
which phonological segmentation exceeds reading. Negative
mean difference scores indicate the extent to which
reading exceeds phonological segmentation.
Links in First and Second Grades
11
Figure Captions
Figure 1 Reading and spelling given segmentation response for
exception words
Figure 2 Reading and spelling given segmentation response for regular
words
1.6
ERIC
V
0
L
0
ID
0
c
E
L
0
L
(I)
(L
"0
L
0
o
3
-p
a)
a
<D
CO
o
<D
u
u
o
u
d
o
0 "8
d
ID
s
ID
c
0
a o
0 fa
X 0
lj g
□
a
a a
□ □
I I I M I J i : 8
a
Q
q a a a m a a
-i — i 1 r~
* I 1
a
c
-rl
r-*
H
0
a
0)
4J
0
a)
L
L
0
0
c
-ri
H
H
0
a
a)
0
L
L
0
0
CM
(I)
TJ
D
L
0
a)
0
c
0
E
L
0
L
0
a.
"0
L
d
o
d
a)
(D
CO
-P
0
(D
O
u
d
o
o 3
d
a)
c S
0 g>
!p co
(D Q)
u.
o
o
d
0
X
JiJ
+
X ♦
+ +
"T r 1 1 i 1 1 1 1— i r r
* +
— I 1
a
v- 0
v §
o &
* ?
■P
O
(1)
O
u
0
(1)
0
c
0
E
L
0
L
<D
Q_
V
L
0
d
o
•H
+)
d
d
a)
6
w
L *»
„ a
^ 8
en jg
ID
a a
□
□
a
Q
I I I!
9 I 3
a a
_ a a
a a
—i 1 1 1 1 1 1 1 r~
I I t «. ■. I I 3 6
a
c
•H
H
H
<D
a
u
I)
L
L
0
O
a
c
M
H
a
tn
-M
U
0
L
L
0
o
id 3
o a
o>
0
u
u
o
o
L
0
(D
0
c
0
E
L
0
L
(D
(L
"0
L
0
d
o
•r-l
<d
d
a)
DO
a)
h °
— u
(1)
c
a.
00
o
o
o
T 1 t r i 1 y r 1 1 1 r~
♦ ♦
c
9
a.
cn
o
4)
O
o
-i 1 r 1 1 1 1 1 1 1 1 1-
ERLC